Research
Original Investigation
Higher Risk of Developing Sudden Sensorineural Hearing Loss in Patients With Chronic Otitis Media Yung-Chang Yen, MD, PhD; Charlene Lin, BS; Shih-Feng Weng, PhD; Yung-Song Lin, MD
IMPORTANCE Several sources have suggested an association between chronic sensory
hearing impairment and chronic otitis media (COM). However, to our knowledge, no studies have evaluated the risk of sudden sensorineural hearing loss (SSNHL) in patients with COM (COM-positive). OBJECTIVE To examine the risk of developing SSNHL in COM-positive patients. DESIGN, SETTING, AND PARTICIPANTS This was a retrospective cohort study; we compared 10 248 patients with newly diagnosed COM from January 1, 2001, through December 31, 2008, with 30 744 age- and sex-matched controls using data from Taiwan’s National Health Insurance Research Database. METHODS We followed each patient and evaluated the incidence of SSNHL. MAIN OUTCOMES AND MEASURES The incidence of SSNHL at the end of 2011. RESULTS The incidence of SSNHL was 3 times higher in the COM-positive cohort than in the COM-negative cohort (14.47 vs 4.83 per 10 000 person-years). Cox proportional hazard regressions showed that the adjusted hazard ratio (AHR) was 3.02 (95% CI, 2.30-3.98). A stratified analysis showed that the highest risk of developing SSNHL was in the first follow-up year (incidence rate ratio [IRR], 3.87; 95% CI, 1.93-7.79). Thereafter, the risk declined during years 1 to 5 and then peaked (IRR, 3.01; 95% CI, 1.89-4.79). Patients who needed surgery had a higher incidence of SSNHL (AHR, 2.69; 95% CI, 1.62-4.48) compared with patients who needed only medication and observation. CONCLUSIONS AND RELEVANCE Chronic otitis media was significantly associated with a higher
risk of developing SSNHL. JAMA Otolaryngol Head Neck Surg. 2015;141(5):429-435. doi:10.1001/jamaoto.2015.102 Published online March 5, 2015.
S
everal disparate sources,1-5 including histopathological and audiological studies, have suggested an association between chronic otitis media (COM) and chronic sensory hearing impairment. It is plausible that COM-mediated cochlear disease is related to chronic sensory hearing impairment.1-5 For example, Cureoglu et al5 discovered a significant loss of outer and inner hair cells in the basal turn of the cochlea in the temporal bones of patients with COM (hereinafter, COM-positive) compared with the control group (COM-negative). Papp et al3 discovered that bone conduction threshold averages were significantly higher in 121 patients with unilateral chronic suppurative otitis media. The pathogenesis of chronic sensory hearing impairment may be related to inflammatory noxious substances that cross the round window membrane, which leads to serous labyrinthitis, or to fluid in the middle ear that impedes oxygen transport to the inner ear, or an adverse effect of ototoxic drugs.6 jamaotolaryngology.com
Author Affiliations: Department of Nursing, Min Hwei College of Health Care Management, Tainan, Taiwan (Yen); Chi Mei Medical Center, LiouYing Campus, Tainan, Taiwan (Yen); Department of Molecular and Cell Biology, University of California, Berkeley (C. Lin); Department of Medical Research, Chi Mei Medical Center, Tainan, Taiwan (Weng); Department of Hospital and Health Care Administration, Chia-Nan University of Pharmacy and Science, Tainan, Taiwan (Weng); Center for General Education, Southern Taiwan University of Science and Technology, Tainan, Taiwan (Y.-S. Lin); Department of Otolaryngology, Chi Mei Medical Center, Tainan, Taiwan (Y.-S. Lin). Corresponding Author: Yung-Song Lin, MD, Department of Otolaryngology, Chi Mei Medical Center, 901 Zhonghua Rd, Yongkang District, Tainan City 710, Taiwan ([email protected]).
To our knowledge, there are, in the English literature, no case series or cohort studies on the risk of sudden sensorineural hearing loss (SSNHL) in patients with COM. When the symptoms of SSNHL appear, they might be startling and unsettling for the patient with COM and also for the physician. There are possibly some similar underlying mechanisms for both SSNHL and chronic hearing impairment. The results of the studies cited1-5 inspired us to study COM as a risk factor of SSNHL. We wanted to test the hypothesis that there is a discernable link between chronic hearing impairment and COMmediated cochlear disease; thus, we chose to examine the association between COM and SSNHL. To identify the effect of COM on the risk of developing SSNHL, we did a population-based cohort study using Taiwan’s National Health Insurance Research Database (NHIRD). We also examined the relationship between the severity of COM and the risk of developing SSNHL.
(Reprinted) JAMA Otolaryngology–Head & Neck Surgery May 2015 Volume 141, Number 5
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archotol.jamanetwork.com/ by a University of Chicago Libraries User on 07/02/2015
429
Research Original Investigation
Sudden Sensorineural Hearing Loss and Chronic Otitis Media
Methods Data Sources Taiwan’s National Health Insurance (NHI) program, a universal and mandatory single-payer system, was launched on March 1, 1995. More than 98% of Taiwan’s 22.96 million legal residents (citizens and noncitizens) are enrolled in this program. The NHIRD is an administrative compilation of patient demographic and clinical data and of monetary claims made by health care providers and paid for by the NHI program. For this study, we used the Longitudinal Health Insurance Database (LHID) 2012, which contains all claims data from 1996 to 2011 of 1 million randomly selected NHI beneficiaries. The database contains encrypted patient identification numbers, the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes of diagnoses, inpatient claims, ambulatory care claims, procedures and prescriptions, dates of admission and discharge, and patient registration data, which include basic sociodemographic information: sex, date of birth, area of residence, date of NHI registration, and date of termination of NHI coverage.
Study Sample There were 2 cohorts in this retrospective cohort study: (1) a COM-positive cohort (patients with COM) and (2) a COMnegative cohort (patients without COM, a matched control [comparison] cohort). COM (ICD-9-CM codes 381.1-381.3; 382.1382.3) and SSNHL (ICD-9-CM code 388.2) were defined if patients had a minimum of 3 outpatient service claims or any single hospitalization with a corresponding diagnosis made by a secondary or tertiary referral hospital. Patients with a diagnosis of labyrinthitis (ICD-9-CM code 386.3) or meningitis (ICD9-CM code 320-322) were excluded. The COM-positive cohort was composed of patients who were newly diagnosed from January 1, 2001, to December 31, 2008, as having COM. The COM-negative control cohort included randomly selected patients without SSNHL-related medical claims but propensityscore matched to the study cohort for the baseline covariates of age, sex, comorbidities, area of residence, and monthly income.7 We defined the index date as the first occurrence of COM for patients in the COM-positive cohort. Matching the year of the index date to the COM-positive cohort, we further defined the index date for the COM-negative cohort. Comorbid major diseases that existed before the index date were diabetes mellitus (DM) (ICD-9-CM code 250), hypertension (HTN) (ICD-9-CM codes 401-405), chronic kidney disease (CKD) (ICD9-CM code 585), and hypercholesterolemia (HCh) (ICD-9-CM codes 272.0). The comorbidities were identified using a diagnosis made during admission or by a specialist during the patient’s hospital visit. The routine peer review system of the Taiwan Health Insurance Bureau looks over the diagnoses made in every referral hospital to ensure that diagnostic methods are standardized. To investigate the association between the occurrence of SSNHL and the pathological progress of COM, only patients with newly diagnosed COM from January 1, 2001, to December 31, 2008, were included. The claims data from 1996 to 2000 were used to confirm that none of the enrolled pa430
tients had ever been diagnosed as having COM or SSNHL before 2001. The cohort in this study consists of patients who have been evaluated in terms of the incidence of SSNHL that developed 3 months beyond the index date through the end of 2011 or until the death of the patients. The institutional review board of Chi Mei Medical Center approved this study.
Statistical Analysis All statistical analyses were performed using SAS software (version 9.2 for Windows; SAS Institute Inc). Differences in sociodemographic characteristics and comorbidities between the COM-positive cohort and COM-negative control cohort were compared using Pearson χ2 tests. We calculated the incidence rate using the following formula: number of SSNHL incidents detected during the follow-up period divided by the total number of person-years for each cohort group; both were classified by sex, age, and years of follow-up. Poisson regression was used to estimate the incidence rate ratio (IRR) of developing SSNHL between the COM-positive and COMnegative cohorts. We used Cox proportional hazard models to assess the risks for developing SSNHL with comorbid DM, HTN, CKD, and HCh. To calculate the total incidence rates of SSNHL in the 2 cohort groups, we used Cox proportional hazard regression analyses and Kaplan-Meier analyses. The differences in the cumulative incidence rates between the 2 cohorts were analyzed using a log-rank test. Each dichotomous variable in the model was tested for proportionality using investigative, diagnostic log-log survival plots. In this study, the follow-up period was time to event (event = SSNHL), time to death, or time to the end of follow-up. When the first event occurred, the end point was the date of the first event. Therefore, in the present study, an IRR is a hazard ratio (HR) with a very particular set of assumptions: the hazard is both proportional and constant. Significance was set at P < .05 (2-tailed). We used Stata statistical software (version 12; STATA Corp) for power calculations. The Stata command stpower logrank can be used to estimate power for survival data for a sample of at least 10 248 patients, an effect size of 3.0 (expressed as an HR), and an α of .05 with a 2-sided test are set in STATA software. The statistical power was estimated to be more than 99% and would be able to detect any significant difference in 2 cohorts.
Results Incidence of SSNHL Relative to Patient Characteristics We enrolled from the NHIRD 10 248 COM-positive patients who, between January 1, 2001, and December 31, 2008, met the eligibility criteria for this study. For the COM-negative cohort, we enrolled 30 744 age- and sex-matched patients (ratio, 1:3) (Table 1). There were no significant baseline differences in the prevalence of DM, CKD, HTN, or HCh, or in the monthly incomes or the areas of residence for the 2 cohorts (Table 1). Patients with newly diagnosed COM had 3 times the incidence of SSNHL than did COM-negative patients (P < .01) (Figure). When the follow-up period was over, the SSNHL
JAMA Otolaryngology–Head & Neck Surgery May 2015 Volume 141, Number 5 (Reprinted)
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archotol.jamanetwork.com/ by a University of Chicago Libraries User on 07/02/2015
jamaotolaryngology.com
Sudden Sensorineural Hearing Loss and Chronic Otitis Media
Original Investigation Research
Table 1. Demographic Characteristics and Comorbidities of Patients With and Without Chronic Otitis Media No. (%) COM-Positive (n = 10 248)
COM-Negative (Controls) (n = 30 744)
0-34
3210 (31.32)
9620 (31.29)
35-49
2467 (24.07)
7376 (23.99)
50-64
2490 (24.30)
7539 (24.52)
≥65
2081 (20.31)
6209 (20.20)
Male
5184 (50.59)
15 431 (50.19)
Female
5064 (49.41)
15 313 (49.81)
P Value
Characteristic Age, y
.97
Sex .49
Baseline Comorbidity DM Yes
845 (8.25)
2501 (8.13)
No
9403 (91.75)
28 243 (91.87)
.72
HTN Yes
1714 (16.73)
5305 (17.26)
No
8534 (83.27)
25 439 (82.74)
.22
CKD Yes No
264 (1.31) 19 894 (98.69)
711 (1.18) 59 763 (98.82)
.14
HCh Yes
576 (5.62)
576 (5.62)
No
9672 (94.38)
9672 (94.38)
.37
Surgery required Yes
672 (6.56)
No
9576 (93.44)
NA
Area of residence in Taiwan North
4925 (48.06)
14 675 (47.73)
Central
1867 (18.22)
5670 (18.44)
South
3206 (31.28)
9685 (31.50)
250 (2.44)
714 (2.32)
East
.81
Monthly income, NT$ 25 001
2103 (20.52)
6360 (20.69)
IRR of the COM-positive cohort was generally significantly (P < .01 for all age groups) higher than that of the COMnegative control cohort. In the COM-positive cohort, patients who were 50 to 64 years old had the highest incidence (19.43 per 10 000 person-years) of SSNHL, and patients who were 35 to 49 years old had the highest IRR, 4.28 (95% CI, 2.487.39) (Table 2).
SSNHL and Its Associated Comorbidities Considering all the stratified analyses we did that are listed with their matched variables, the COM-positive cohort was associated with a higher risk of SSNHL. Because of the differential effect of COM stratified for some of these matching variables, the matching had to be retained in the analyses. We used a multivariate Cox proportional hazards regression analysis to show that there was an adjusted HR (AHR) for SSNHL of 3.02 (95% CI, 2.30-3.98) for COM (Table 3). In genjamaotolaryngology.com
.92
Abbreviations: CKD, chronic kidney disease; COM, chronic otitis media; DM, diabetes mellitus; HCh, hypercholesterolemia; HTN, hypertension; NA, not applicable; NT$, New Taiwan dollars (US$1 = NT$30).
eral, the HR increased with age. Sex, however, was not a significant risk for SSNHL. Table 3 indicates that none of the comorbid diseases (DM, HTN, CKD, and HCh) were overall risk factors for SSNHL. Furthermore, we calculated the IRR for each comorbidity between the COM-positive and COM-negative control cohorts to examine the effects of the comorbidities (Table 2). In the comorbid HTN and HCh subgroups, the incidence of ISSNHL between the COM-positive and COM-negative control cohorts was significantly different (IRR, 1.91; 95% CI, 1.06-3.42 vs 2.91; 95% CI, 1.02-8.29), respectively. Nevertheless, both HTN and HCh created an attenuation effect for the IRR value of COM-positive to COM-negative control cohorts, that is, the IRR measured for the overall study was 3.00, but the IRR measured for the HTN comorbidity subgroup was only 1.91, and that for the HCh comorbidity subgroup was 2.91.
(Reprinted) JAMA Otolaryngology–Head & Neck Surgery May 2015 Volume 141, Number 5
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archotol.jamanetwork.com/ by a University of Chicago Libraries User on 07/02/2015
431
Research Original Investigation
Sudden Sensorineural Hearing Loss and Chronic Otitis Media
Requirements for Surgical Treatment Relative to the Risk of Developing SSNHL We also evaluated the association between the requirement for surgical treatment, either tympanoplasty and mastoidectomy, and the risk of SSNHL (Table 4). Patients who required surgical treatment had a higher incidence of SSNHL (AHR, 2.69; 95% CI, 1.62-4.48) than did patients who required only medication and observation, after sex, age, comorbidities, area of residence, and monthly income had
been adjusted for. There were 672 patients in the COMpositive cohort who needed surgery. Before surgery, 9 of them developed SSNHL. Another 10 patients developed SSNHL within an average of 3.12 years after COM surgery (range, 0.96-5.58 years). There was almost no association between the occurrence of SSNHL and the surgical treatments in our studied cohorts.
Discussion Figure. The Cumulative Incidence Rate of Sudden Sensorineural Hearing Loss (SSNHL) in Taiwanese Patients With Chronic Otitis Media (COM) 0.02
Cumulative Incidence Rate
COM Controls
Log rank P < .01 0.01
0 0
2
4
6
8
10
3818 11 879
1168 3702
12
Time, y
Number at risk: COM: 10 248 Controls: 30 744
10 012 30 361
8509 25 971
6358 19 561
Patients with newly diagnosed COM had higher incidence of SSNHL than did patients without COM (P < .01).
Higher Risk of SSNHL in Patients With COM We found that the patients with newly diagnosed COM had a greater chance of developing SSNHL than did the COMnegative controls. This is, to the best of our knowledge, the first nationwide, retrospective cohort study that examines the risk of SSNHL in COM-positive patients. It is noteworthy that 17.82% of the COM-positive patients who later developed SSNHL did so within the first year after the COM diagnosis, 24.75% with 1 to 2 years after, 22.77% within 3 to 4 years after, and 34.65% more than 5 years after. Nearly 70% of the COM-positive patients who later developed SSNHL did so within 4 years after the COM diagnosis. The age group with the largest number of patients who developed SSNHL was the 50- to 64-year-old group (n = 32), and the group with the second largest number was the 35- to 49-year-old group (n = 31). A timely effort to detect SSNHL is recommended, especially within 4 years after the initial COM diagnosis.
Table 2. Risk of SSNHL for Patients With and Without (Controls) Chronic Otitis Media (COM) COM-Positive
COM-Negative (Controls)
No.
SSNHL
PY
Ratea
No.
SSNHL
PY
Ratea
IRR (95% CI)
P Value
10 248
101
69 785.95
14.47
30 744
103
213 181.48
4.83
3.00 (2.28-3.94)
Original Investigation
Higher Risk of Developing Sudden Sensorineural Hearing Loss in Patients With Chronic Otitis Media Yung-Chang Yen, MD, PhD; Charlene Lin, BS; Shih-Feng Weng, PhD; Yung-Song Lin, MD
IMPORTANCE Several sources have suggested an association between chronic sensory
hearing impairment and chronic otitis media (COM). However, to our knowledge, no studies have evaluated the risk of sudden sensorineural hearing loss (SSNHL) in patients with COM (COM-positive). OBJECTIVE To examine the risk of developing SSNHL in COM-positive patients. DESIGN, SETTING, AND PARTICIPANTS This was a retrospective cohort study; we compared 10 248 patients with newly diagnosed COM from January 1, 2001, through December 31, 2008, with 30 744 age- and sex-matched controls using data from Taiwan’s National Health Insurance Research Database. METHODS We followed each patient and evaluated the incidence of SSNHL. MAIN OUTCOMES AND MEASURES The incidence of SSNHL at the end of 2011. RESULTS The incidence of SSNHL was 3 times higher in the COM-positive cohort than in the COM-negative cohort (14.47 vs 4.83 per 10 000 person-years). Cox proportional hazard regressions showed that the adjusted hazard ratio (AHR) was 3.02 (95% CI, 2.30-3.98). A stratified analysis showed that the highest risk of developing SSNHL was in the first follow-up year (incidence rate ratio [IRR], 3.87; 95% CI, 1.93-7.79). Thereafter, the risk declined during years 1 to 5 and then peaked (IRR, 3.01; 95% CI, 1.89-4.79). Patients who needed surgery had a higher incidence of SSNHL (AHR, 2.69; 95% CI, 1.62-4.48) compared with patients who needed only medication and observation. CONCLUSIONS AND RELEVANCE Chronic otitis media was significantly associated with a higher
risk of developing SSNHL. JAMA Otolaryngol Head Neck Surg. 2015;141(5):429-435. doi:10.1001/jamaoto.2015.102 Published online March 5, 2015.
S
everal disparate sources,1-5 including histopathological and audiological studies, have suggested an association between chronic otitis media (COM) and chronic sensory hearing impairment. It is plausible that COM-mediated cochlear disease is related to chronic sensory hearing impairment.1-5 For example, Cureoglu et al5 discovered a significant loss of outer and inner hair cells in the basal turn of the cochlea in the temporal bones of patients with COM (hereinafter, COM-positive) compared with the control group (COM-negative). Papp et al3 discovered that bone conduction threshold averages were significantly higher in 121 patients with unilateral chronic suppurative otitis media. The pathogenesis of chronic sensory hearing impairment may be related to inflammatory noxious substances that cross the round window membrane, which leads to serous labyrinthitis, or to fluid in the middle ear that impedes oxygen transport to the inner ear, or an adverse effect of ototoxic drugs.6 jamaotolaryngology.com
Author Affiliations: Department of Nursing, Min Hwei College of Health Care Management, Tainan, Taiwan (Yen); Chi Mei Medical Center, LiouYing Campus, Tainan, Taiwan (Yen); Department of Molecular and Cell Biology, University of California, Berkeley (C. Lin); Department of Medical Research, Chi Mei Medical Center, Tainan, Taiwan (Weng); Department of Hospital and Health Care Administration, Chia-Nan University of Pharmacy and Science, Tainan, Taiwan (Weng); Center for General Education, Southern Taiwan University of Science and Technology, Tainan, Taiwan (Y.-S. Lin); Department of Otolaryngology, Chi Mei Medical Center, Tainan, Taiwan (Y.-S. Lin). Corresponding Author: Yung-Song Lin, MD, Department of Otolaryngology, Chi Mei Medical Center, 901 Zhonghua Rd, Yongkang District, Tainan City 710, Taiwan ([email protected]).
To our knowledge, there are, in the English literature, no case series or cohort studies on the risk of sudden sensorineural hearing loss (SSNHL) in patients with COM. When the symptoms of SSNHL appear, they might be startling and unsettling for the patient with COM and also for the physician. There are possibly some similar underlying mechanisms for both SSNHL and chronic hearing impairment. The results of the studies cited1-5 inspired us to study COM as a risk factor of SSNHL. We wanted to test the hypothesis that there is a discernable link between chronic hearing impairment and COMmediated cochlear disease; thus, we chose to examine the association between COM and SSNHL. To identify the effect of COM on the risk of developing SSNHL, we did a population-based cohort study using Taiwan’s National Health Insurance Research Database (NHIRD). We also examined the relationship between the severity of COM and the risk of developing SSNHL.
(Reprinted) JAMA Otolaryngology–Head & Neck Surgery May 2015 Volume 141, Number 5
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archotol.jamanetwork.com/ by a University of Chicago Libraries User on 07/02/2015
429
Research Original Investigation
Sudden Sensorineural Hearing Loss and Chronic Otitis Media
Methods Data Sources Taiwan’s National Health Insurance (NHI) program, a universal and mandatory single-payer system, was launched on March 1, 1995. More than 98% of Taiwan’s 22.96 million legal residents (citizens and noncitizens) are enrolled in this program. The NHIRD is an administrative compilation of patient demographic and clinical data and of monetary claims made by health care providers and paid for by the NHI program. For this study, we used the Longitudinal Health Insurance Database (LHID) 2012, which contains all claims data from 1996 to 2011 of 1 million randomly selected NHI beneficiaries. The database contains encrypted patient identification numbers, the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes of diagnoses, inpatient claims, ambulatory care claims, procedures and prescriptions, dates of admission and discharge, and patient registration data, which include basic sociodemographic information: sex, date of birth, area of residence, date of NHI registration, and date of termination of NHI coverage.
Study Sample There were 2 cohorts in this retrospective cohort study: (1) a COM-positive cohort (patients with COM) and (2) a COMnegative cohort (patients without COM, a matched control [comparison] cohort). COM (ICD-9-CM codes 381.1-381.3; 382.1382.3) and SSNHL (ICD-9-CM code 388.2) were defined if patients had a minimum of 3 outpatient service claims or any single hospitalization with a corresponding diagnosis made by a secondary or tertiary referral hospital. Patients with a diagnosis of labyrinthitis (ICD-9-CM code 386.3) or meningitis (ICD9-CM code 320-322) were excluded. The COM-positive cohort was composed of patients who were newly diagnosed from January 1, 2001, to December 31, 2008, as having COM. The COM-negative control cohort included randomly selected patients without SSNHL-related medical claims but propensityscore matched to the study cohort for the baseline covariates of age, sex, comorbidities, area of residence, and monthly income.7 We defined the index date as the first occurrence of COM for patients in the COM-positive cohort. Matching the year of the index date to the COM-positive cohort, we further defined the index date for the COM-negative cohort. Comorbid major diseases that existed before the index date were diabetes mellitus (DM) (ICD-9-CM code 250), hypertension (HTN) (ICD-9-CM codes 401-405), chronic kidney disease (CKD) (ICD9-CM code 585), and hypercholesterolemia (HCh) (ICD-9-CM codes 272.0). The comorbidities were identified using a diagnosis made during admission or by a specialist during the patient’s hospital visit. The routine peer review system of the Taiwan Health Insurance Bureau looks over the diagnoses made in every referral hospital to ensure that diagnostic methods are standardized. To investigate the association between the occurrence of SSNHL and the pathological progress of COM, only patients with newly diagnosed COM from January 1, 2001, to December 31, 2008, were included. The claims data from 1996 to 2000 were used to confirm that none of the enrolled pa430
tients had ever been diagnosed as having COM or SSNHL before 2001. The cohort in this study consists of patients who have been evaluated in terms of the incidence of SSNHL that developed 3 months beyond the index date through the end of 2011 or until the death of the patients. The institutional review board of Chi Mei Medical Center approved this study.
Statistical Analysis All statistical analyses were performed using SAS software (version 9.2 for Windows; SAS Institute Inc). Differences in sociodemographic characteristics and comorbidities between the COM-positive cohort and COM-negative control cohort were compared using Pearson χ2 tests. We calculated the incidence rate using the following formula: number of SSNHL incidents detected during the follow-up period divided by the total number of person-years for each cohort group; both were classified by sex, age, and years of follow-up. Poisson regression was used to estimate the incidence rate ratio (IRR) of developing SSNHL between the COM-positive and COMnegative cohorts. We used Cox proportional hazard models to assess the risks for developing SSNHL with comorbid DM, HTN, CKD, and HCh. To calculate the total incidence rates of SSNHL in the 2 cohort groups, we used Cox proportional hazard regression analyses and Kaplan-Meier analyses. The differences in the cumulative incidence rates between the 2 cohorts were analyzed using a log-rank test. Each dichotomous variable in the model was tested for proportionality using investigative, diagnostic log-log survival plots. In this study, the follow-up period was time to event (event = SSNHL), time to death, or time to the end of follow-up. When the first event occurred, the end point was the date of the first event. Therefore, in the present study, an IRR is a hazard ratio (HR) with a very particular set of assumptions: the hazard is both proportional and constant. Significance was set at P < .05 (2-tailed). We used Stata statistical software (version 12; STATA Corp) for power calculations. The Stata command stpower logrank can be used to estimate power for survival data for a sample of at least 10 248 patients, an effect size of 3.0 (expressed as an HR), and an α of .05 with a 2-sided test are set in STATA software. The statistical power was estimated to be more than 99% and would be able to detect any significant difference in 2 cohorts.
Results Incidence of SSNHL Relative to Patient Characteristics We enrolled from the NHIRD 10 248 COM-positive patients who, between January 1, 2001, and December 31, 2008, met the eligibility criteria for this study. For the COM-negative cohort, we enrolled 30 744 age- and sex-matched patients (ratio, 1:3) (Table 1). There were no significant baseline differences in the prevalence of DM, CKD, HTN, or HCh, or in the monthly incomes or the areas of residence for the 2 cohorts (Table 1). Patients with newly diagnosed COM had 3 times the incidence of SSNHL than did COM-negative patients (P < .01) (Figure). When the follow-up period was over, the SSNHL
JAMA Otolaryngology–Head & Neck Surgery May 2015 Volume 141, Number 5 (Reprinted)
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archotol.jamanetwork.com/ by a University of Chicago Libraries User on 07/02/2015
jamaotolaryngology.com
Sudden Sensorineural Hearing Loss and Chronic Otitis Media
Original Investigation Research
Table 1. Demographic Characteristics and Comorbidities of Patients With and Without Chronic Otitis Media No. (%) COM-Positive (n = 10 248)
COM-Negative (Controls) (n = 30 744)
0-34
3210 (31.32)
9620 (31.29)
35-49
2467 (24.07)
7376 (23.99)
50-64
2490 (24.30)
7539 (24.52)
≥65
2081 (20.31)
6209 (20.20)
Male
5184 (50.59)
15 431 (50.19)
Female
5064 (49.41)
15 313 (49.81)
P Value
Characteristic Age, y
.97
Sex .49
Baseline Comorbidity DM Yes
845 (8.25)
2501 (8.13)
No
9403 (91.75)
28 243 (91.87)
.72
HTN Yes
1714 (16.73)
5305 (17.26)
No
8534 (83.27)
25 439 (82.74)
.22
CKD Yes No
264 (1.31) 19 894 (98.69)
711 (1.18) 59 763 (98.82)
.14
HCh Yes
576 (5.62)
576 (5.62)
No
9672 (94.38)
9672 (94.38)
.37
Surgery required Yes
672 (6.56)
No
9576 (93.44)
NA
Area of residence in Taiwan North
4925 (48.06)
14 675 (47.73)
Central
1867 (18.22)
5670 (18.44)
South
3206 (31.28)
9685 (31.50)
250 (2.44)
714 (2.32)
East
.81
Monthly income, NT$ 25 001
2103 (20.52)
6360 (20.69)
IRR of the COM-positive cohort was generally significantly (P < .01 for all age groups) higher than that of the COMnegative control cohort. In the COM-positive cohort, patients who were 50 to 64 years old had the highest incidence (19.43 per 10 000 person-years) of SSNHL, and patients who were 35 to 49 years old had the highest IRR, 4.28 (95% CI, 2.487.39) (Table 2).
SSNHL and Its Associated Comorbidities Considering all the stratified analyses we did that are listed with their matched variables, the COM-positive cohort was associated with a higher risk of SSNHL. Because of the differential effect of COM stratified for some of these matching variables, the matching had to be retained in the analyses. We used a multivariate Cox proportional hazards regression analysis to show that there was an adjusted HR (AHR) for SSNHL of 3.02 (95% CI, 2.30-3.98) for COM (Table 3). In genjamaotolaryngology.com
.92
Abbreviations: CKD, chronic kidney disease; COM, chronic otitis media; DM, diabetes mellitus; HCh, hypercholesterolemia; HTN, hypertension; NA, not applicable; NT$, New Taiwan dollars (US$1 = NT$30).
eral, the HR increased with age. Sex, however, was not a significant risk for SSNHL. Table 3 indicates that none of the comorbid diseases (DM, HTN, CKD, and HCh) were overall risk factors for SSNHL. Furthermore, we calculated the IRR for each comorbidity between the COM-positive and COM-negative control cohorts to examine the effects of the comorbidities (Table 2). In the comorbid HTN and HCh subgroups, the incidence of ISSNHL between the COM-positive and COM-negative control cohorts was significantly different (IRR, 1.91; 95% CI, 1.06-3.42 vs 2.91; 95% CI, 1.02-8.29), respectively. Nevertheless, both HTN and HCh created an attenuation effect for the IRR value of COM-positive to COM-negative control cohorts, that is, the IRR measured for the overall study was 3.00, but the IRR measured for the HTN comorbidity subgroup was only 1.91, and that for the HCh comorbidity subgroup was 2.91.
(Reprinted) JAMA Otolaryngology–Head & Neck Surgery May 2015 Volume 141, Number 5
Copyright 2015 American Medical Association. All rights reserved.
Downloaded From: http://archotol.jamanetwork.com/ by a University of Chicago Libraries User on 07/02/2015
431
Research Original Investigation
Sudden Sensorineural Hearing Loss and Chronic Otitis Media
Requirements for Surgical Treatment Relative to the Risk of Developing SSNHL We also evaluated the association between the requirement for surgical treatment, either tympanoplasty and mastoidectomy, and the risk of SSNHL (Table 4). Patients who required surgical treatment had a higher incidence of SSNHL (AHR, 2.69; 95% CI, 1.62-4.48) than did patients who required only medication and observation, after sex, age, comorbidities, area of residence, and monthly income had
been adjusted for. There were 672 patients in the COMpositive cohort who needed surgery. Before surgery, 9 of them developed SSNHL. Another 10 patients developed SSNHL within an average of 3.12 years after COM surgery (range, 0.96-5.58 years). There was almost no association between the occurrence of SSNHL and the surgical treatments in our studied cohorts.
Discussion Figure. The Cumulative Incidence Rate of Sudden Sensorineural Hearing Loss (SSNHL) in Taiwanese Patients With Chronic Otitis Media (COM) 0.02
Cumulative Incidence Rate
COM Controls
Log rank P < .01 0.01
0 0
2
4
6
8
10
3818 11 879
1168 3702
12
Time, y
Number at risk: COM: 10 248 Controls: 30 744
10 012 30 361
8509 25 971
6358 19 561
Patients with newly diagnosed COM had higher incidence of SSNHL than did patients without COM (P < .01).
Higher Risk of SSNHL in Patients With COM We found that the patients with newly diagnosed COM had a greater chance of developing SSNHL than did the COMnegative controls. This is, to the best of our knowledge, the first nationwide, retrospective cohort study that examines the risk of SSNHL in COM-positive patients. It is noteworthy that 17.82% of the COM-positive patients who later developed SSNHL did so within the first year after the COM diagnosis, 24.75% with 1 to 2 years after, 22.77% within 3 to 4 years after, and 34.65% more than 5 years after. Nearly 70% of the COM-positive patients who later developed SSNHL did so within 4 years after the COM diagnosis. The age group with the largest number of patients who developed SSNHL was the 50- to 64-year-old group (n = 32), and the group with the second largest number was the 35- to 49-year-old group (n = 31). A timely effort to detect SSNHL is recommended, especially within 4 years after the initial COM diagnosis.
Table 2. Risk of SSNHL for Patients With and Without (Controls) Chronic Otitis Media (COM) COM-Positive
COM-Negative (Controls)
No.
SSNHL
PY
Ratea
No.
SSNHL
PY
Ratea
IRR (95% CI)
P Value
10 248
101
69 785.95
14.47
30 744
103
213 181.48
4.83
3.00 (2.28-3.94)
